To establish medically useful guidelines for analytic precision of commonly used clinical laboratory procedures, the authors conducted a mail survey of physicians selected randomly from national lists of specialists. They were asked to review, briefly, outlined clinical problems and select the change in test results that would alter their diagnosis or treatment or prompt further assessment of the patient's condition. The responses were used to calculate goals for laboratory precision that are sufficient to meet the present requirements of the average physician. The medically useful limits were compared with existing performance of laboratories as reported in national proficiency surveys and regional quality control programs. Almost all the common laboratory procedures they studied are being assayed at a precision level adequate for the perceived needs. The authors suggest that, for the more common constituents, further progress in clinical laboratory testing will depend on factors other than the improvement of analytic precision.